 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of

William R. Wade, the author of this book.

Written by Gi-uk Kim
The Solutions of Analysis

Writer : Gi-uk Kim.

Belongs To : Department of Mathematics,
SOGANG University,
1 Shinsoo-dong, Mapo-gu, Seoul,
in the Republic of Korea.

No part of this document may be reproduced, in any form or by any

means, without permission in writing from this writer.
And if you find errors, I would appreciate your contacting me at the
e-mail address below.
My E-mail Address : [email protected]
My Cellular Phone Number : 010-2289-5204
These solutions might be inexact, yet do you rely on them?
Copyright this writer. All rights reserved.

Chapter 2 Sequences in R
2.1 Limits of Sequences

p.38
Exercises
1. Using the method of Example 2.2, prove that the following limits exist.
1 → 3 as n → ∞.
(a) 3 +
n
Proof) Let ε > 0 be given. By the Archimedean Principle,
1
we can choose an N ∈ N such that N > . Thus, we see that
ε

n ≥ N implies ∣( 3 + 1n ) - 3∣= 1 ≤ 1
n N
< ε.

(b) 2 ( 1 - 1n ) → 2 as n → ∞.

Proof) Let ε > 0 be given. By the Archimedean Principle,

2
we can choose an N ∈ N such that N > . Thus, we see that
ε

n ≥ N implies ∣2 ( 1 - 1n ) - 2∣= 2
n
≤
2
N
< ε.

5+n
(c) → 0 as n → ∞.
n2
Proof) Let ε > 0 be given. By the Archimedean Principle,

These solutions might be inexact, yet do you rely on them?

- 1 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
6
we can choose an N ∈ N such that N > .
ε
Thus, we see that n ≥ N implies

∣ 5+n
n2
-0 ∣= 5
+
n2 n
1 ≤ 5
+
N2 N
1 ≤ 5
+
N N
1
=
6
N
< ε.

3
(d) π - → π as n → ∞.
n
Proof) Let ε > 0 be given. By Archimedean Principle,
9
we can choose an N ∈ N such that N > .
ε2
Thus, we see that n ≥ N implies

∣( π - 3
n ) - π ∣= 3
n
≤
3
N
< ε.

2. Suppose that x n is a sequence of real numbers that converges to 1 as

n → ∞. Using Definition 2.1, prove that each of the following limits exists.
(a) 1 - x n → 0 as n → ∞.
Proof) Suppose that x n converges to 1 as n → ∞.
By definition, given ε > 0, there is N such that
n ≥ N implies ∣x n - 1 ∣< ε.
Then, n ≥ N implies
∣ ( 1 - x n ) - 0 ∣= ∣-1 ∣∣ x n - 1 ∣ = ∣ x n - 1 ∣ < ε.

(b) 3 x n + 1 → 4 as n → ∞.
Proof) Suppose that x n converges to 1 as n → ∞.

By definition, given ε > 0, there is N such that

ε
n ≥ N implies ∣x n - 1∣< .
3
Then, n ≥ N implies
ε
∣ ( 3 x n + 1 ) - 4 ∣= 3∣ x n - 1 ∣< 3⋅ = ε.
3
2 + xn2
(c) → 3 as n → ∞.
xn
Proof) Suppose that x n converges to 1 as n → ∞.
By definition, given ε > 0, there is N 1 such that
ε
n ≥ N 1 implies ∣x n - 1∣< .
3
1
And, let ε = , then there is N 2 such that
2

These solutions might be inexact, yet do you rely on them?

- 2 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
1
n ≥ N 2 implies ∣x n - 1∣< .
2
1 3 3 1
It follows that < xn < and - < xn - 2 < - .
2 2 2 2

That is, ∣ x1 ∣ < 2 and

n
∣x n - 2∣ <
3
2
.

Let N = max { N 1, N 2 }, then n ≥ N implies

∣ 2 + x n2
xn
-3 ∣ ∣=
2 + x n2 - 3 x n
xn ∣
∣x n - 2∣∣x n - 1∣
=
∣x n∣
3 ⋅2 ⋅∣
< x n - 1∣
2
3 ε
< ⋅2 ⋅
2 3
= ε

3. (a) Prove that { (-1) n } has some subsequences that converge and others that
do not converge.
Proof) Set n k = 2 k, then { (-1) n } has a subsequence { (-1)2k }
that converges to 1 immediately.
Also, set n k = k, then { (-1)n } has a subsequence { (-1)k } (itself)
that does not converge at all.
3n
(b) Find a convergent subsequence of n + ( -1 ) n.
3n
Solution) Let x n = n + (-1 ) n.
Set n k = 2 k + 1, then
3 ( 2k + 1 )
x n k = ( 2 k +1) + (-1 ) ( 2 k +1 )
6k 3
= ( 2 k + 1 ) + (-1 ) (-1 ) ( 2k + 1 )

= ( 2 k + 1 ) + 1 (-1 ) ( 2 k + 1 )

= 0
converges to 0 immediately.

4. (a) Suppose that { bn } is a sequence of nonnegative numbers that converges to

0, and { xn } is a real sequence that satisfies ∣x n - a ∣≤ b n for large n. Prove
that x n converges to a.
Proof) By hypothesis,
given ε > 0, there is N 1 ∈ N such that n ≥ N 1 implies

These solutions might be inexact, yet do you rely on them?

- 3 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
∣b n - 0 ∣< ε.

And, there is N 2 ∈ N such that n ≥ N 2 implies

∣x n - a ∣≤ b n

Let N = max { N 1, N 2 }. Since b n ≥ 0 for all n ∈ N, then

for every ε > 0, n ≥ N implies
∣x n - a ∣≤ b n = ∣b n∣ < ε.
Hence, x n converges to a.
(b) What happens to part (a) if " ≤ b n" is replaced by " ≤ C b n" for some fixed

positive constant C ?
ε
Solution) In part (a), replace ε > 0 by > 0.
C

5. Suppose that x n ∈ R.
(a) Prove that { xn } is bounded if and only if there is a C > 0 such that

∣x n∣≤ C for all n ∈ N.

Proof) (⇒) By Definition 2.7, { xn } is bounded
if and only if { xn } is bounded both above and below.

Then, there are M, m ∈ R such that x n ≤ M and x n ≥ m.

And, x n ≤ M ≤∣M∣ and x n ≥ m ≥ - ∣ m∣.
Set C ' = max { ∣M ∣,∣m ∣}, then C ' ≥ 0 (verify!) and
x n ≤ C ' and x n ≥ - C ' .
By Theorem 1.6, ∣x n∣≤ C ' .
Then, there is ε0 > 0 such that C : = C ' + ε 0 and C > 0.
Hence, ∣x n∣≤ C.

(⇐) If there is a C > 0 such that ∣x n∣≤ C for all n ∈ N,

then - C ≤ x n ≤ C.
Since x n ≤ C and x n ≥ - C, then
{ xn } is bounded both above and below.
xn
(b) Suppose that { xn } is bounded. Prove that → 0 as n → ∞, for all
nk
k ∈ N.
Proof) Suppose that { xn } is bounded.
By 2.1Exercises 5 (a), there is a C > 0 such that
∣x n∣≤ C for all n ∈ N.

These solutions might be inexact, yet do you rely on them?

- 4 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Given ε > 0, there is N ∈ N such that

∣ nx n
k -0 ∣= ∣ xn ∣
∣n k∣
≤ C ≤ C
n k n
for n ≥ N.

1
Since → 0 as n → ∞ (Example 2.2) and 2.1Exercises 4 (b) holds,
n
xn
hence → 0 as n → ∞.
nk

6. (a) Suppose that { xn } and { yn } converge to the same point. Prove that
x n - y n → 0 as n → ∞.

Proof) Let a ∈ R.
Suppose that { xn } and { yn } converge to a.

Then, by definition, for given ε > 0, there are N 1, N 2 ∈ N such that

ε ε
∣x n - a ∣< and ∣y n - a ∣< .
2 2
Set N = max { N 1, N 2 }, then n ≥ N implies
∣ ( xn - yn ) - 0 ∣ = ∣xn - a + a - yn∣

≤∣ x n - a ∣+∣ a - y n ∣

= ∣ x n - a ∣+∣ y n - a ∣

ε ε
< + = ε
2 2
Hence, x n - y n → 0 as n → ∞.
(b) Prove that the sequence {n } does not converge.
Proof) There are two different proofs.
i ) The one way.
Let ε > 0 be given.
Suppose to contrary that
{n } converges to a for some a ∈ R.
By Archimedean Principle, we can choose N ∈ N such that
N > ∣a ∣+ 1.
On the other hand, we see that n ≥ N implies ∣n - a ∣ < ε.
Let ε = 1. By Triangle Inequalities, n ≥ N implies
N - ∣a ∣≤ n - ∣a ∣≤∣n - a ∣ < 1.
That is, N < 1 +∣a ∣.
Thus, N < 1 +∣a ∣< N.
That is, N < N, a contradiction.

These solutions might be inexact, yet do you rely on them?

- 5 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
ii ) The other way.
Suppose to contrary that
{n } converges to a for some a ∈ R.
Then, by Theorem 2.8, {n } is bounded by some M ∈ R.
That is, n ≤ M for all n ∈ N.
Set E : = { n ∈ N∣n ≤ M for some M ∈ R }
Then, E = N.
Since n ≥ 1, then M ≥ 1.
Then, by Remark 1.25, there is an n 0 ∈ N such that
n 0 ≤ M ≤ n 0 + 1 < n 0 + 2.

Use Remark 1.1 (ii ) twice, then n 0 + 2 ∈ N.

But, since M < n 0 + 2, then n 0 + 2 ∉ E, a contradiction.
(c) Show that the converse of part (a) is false.
1
Showing) Set x n = n + and y n = n. Then
n
1
xn - yn = → 0, but x n and y n do not converge.
n

7. (a) Let a be a fixed real number and define x n : = a for n ∈ N. Prove that
the "constant" sequence x n converges.
(b) What does { xn } converge to?
Proof) Let ε > 0 be given.
We assume that x n converges to a.
By Definition 2.1, for every ε > 0,
we can choose an N ∈ N such that
∣x n - a ∣ = ∣a - a ∣ = 0 < ε for n ≥ N.

8. Suppose that { xn } is a sequence in R. Prove that x n converges to a if and

only if EVERY subsequence of x n also converges to a.
Proof) (⇒) Let ε > 0 be given. By hypothesis,
we can choose N ∈ N such that n ≥ N implies ∣x n - a ∣< ε.
By Definition 2.5, n k ∈ N and n 1 < n 2 < n 3 < ⋯, and

n k ≥ k for all k ∈ N. Thus, k ≥ N implies ∣x n k - a ∣< ε.

(⇐) Suppose to contrary that every subsequence of x n converges to a
but x n does not converge to a.

These solutions might be inexact, yet do you rely on them?

- 6 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Then, there are two cases. First, if x n converges to b ( =
/ a ),
by (⇒), every subsequence of x n also converges to b, a contradiction.
Otherwise, if x n does not converge,
since x n is also a subsequence of x n (its own self),
then some subsequences do not converge.
It contradicts that EVERY subsequence of x n converges.

2.2 Limit Theorems

p.43
Exercises
1. Prove that each of the following sequences converges to zero.
n4 + n + 1
sin
n2 + 1
(a) x n = .
n
Proof) We assume that x n converges to zero.

For all n ∈ N, then ∣ sin n4 + n + 1

n2 + 1 ∣≤ 1.
1
By Archimedean Principle, there is an N ∈ N such that N > .
ε
Then, for given ε > 0, n ≥ N implies

∣ x n - 0 ∣=
∣ sin n4 + n + 1
n2 + 1 ∣≤ 1 ≤ 1
< ε.
n n N
n
(b) x n = .
n2 + 1
Proof) We assume that x n converges to zero.

1
By Archimedean Principle, there is an N ∈ N such that N > .
ε
Then, for given ε > 0, n ≥ N implies
n n 1 ≤ 1
∣x n - 0∣= < = < ε.
n2 + 1 n2 n N
2n + 1
(c) x n = .
n +1
Proof) We assume that x n converges to zero.
9
By Archimedean Principle, there is an N ∈ N such that N > .
ε2
Then, for given ε > 0, n ≥ N implies

These solutions might be inexact, yet do you rely on them?

- 7 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
2n + 1
∣x n - 0∣ =
n +1
2n +1 2 1
< = +
n n n
2 1 3
< + =
n n n

≤ 3
N
< ε
n
(d) x n = .
2n
Proof) We assume that x n converges to zero.
Then, by 1.2Exercises 6 (b),
1
n ≤ n n
xn = = .
2n n2 - 1 1-
1
n2
n
And xn = > 0.
2n
1 1
Since → 0 and → 0 as n → ∞, thus by Theorem 2.9 and 2.12,
n n2
n ≤ 0
0 ≤ lim x n = lim = 0.
n →∞ n →∞ 2n 1-0
n → 0 as n → ∞.
Hence,
2n

2. Find the limit (if it exists) of each of the following sequences.

1 + n - 3n2
(a) x n = .
3 - 2n + n2
1 1 → 0 as n → ∞, thus by Theorem 2.12,
Solution) Since → 0 and
n n2
1 + n - 3n2
lim x n = lim
n →∞ n →∞ 3 - 2n + n2
1 1
+ -3
n2 n
= lim
n →∞ 3 2
- +1
n2 n
0+0-3
=
0-0+1
= -3
n3 + n - 5
(b) x n = .
5n3 + n - 1
1 1 → 0 as n → ∞, thus by Theorem 2.12,
Solution) Since → 0 and
n2 n3

These solutions might be inexact, yet do you rely on them?

- 8 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
n3 + n - 5
lim x n = lim
n →∞ n →∞ 5n3 + n - 1
1 5
2 - 1+
n n3
= lim
n →∞ 1 1
5+ 2 - 3
n n
1+0-0
=
5+0-0
1
=
5

2n2 - 1
(c) x n = .
n+1
1 1
Solution) Since → 0 and → 0 as n → ∞, thus by Theorem 2.12,
n n2
2n 2 - 1
lim x n = lim
n →∞ n →∞ n +1
1
2-
n2
= lim
n →∞ 1
1+
n
2-0
=
1+0
= 2
(d) x n = n+1 - n.
1
Solution) We need to show that → 0 as n → ∞.
n
1
By Archimedean Principle, we can choose N ∈ N such that N > .
ε2

Then, n ≥ N implies ∣ 1
n
-0 ≤∣ 1
N
< ε

1 → 0 as n → ∞, thus by Theorem 2.12,

And, since
n
lim x n = lim n+1 - n
n →∞ n →∞

( n+1 - n )( n +1 + n )
= lim
n →∞ n +1 + n
n +1 -n
= lim
n →∞ n+1 + n
1
n
= lim
n →∞ 1
1+ +1
n
0
=
1 +1
= 0

These solutions might be inexact, yet do you rely on them?

- 9 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
3. Prove Theorem 2.12 ( iv ).
( iv ) Suppose that { xn } and { yn } are real sequences. If { xn } and { yn } are

convergent, and in addition, if y n =

/ 0 and lim y n =
/ 0,
n →∞

xn lim x n
n →∞
then lim = .
n →∞ yn lim y n
n →∞

Proof) Suppose that x n and y n converge to x and y (=

/ 0 ).
Since x n converges, by Theorem 2.8 and 2.1Exercises 5 (a),

there is a M > 0 such that ∣x n ∣≤ M for all n ∈ N.

And, since y n converges to y, for given ε > 0,

we can choose N 0 ∈ N such that n ≥ N 0 implies

∣y n - y ∣< ε.
∣y∣
Set ε = > 0, then by Triangle Inequalities
2
∣y∣ ∣y∣
∣y n - y ∣< implies ∣y n∣- ∣y ∣≤∣y n - y ∣ <
2 2
∣y∣
and ∣y ∣- ∣y n∣≤∣y n - y ∣ < .
2
∣y ∣ 3∣y ∣
Thus < ∣y n∣ < .
2 2
1 2
That is, < .
∣y n∣ ∣y ∣

And, by hypothesis, we can choose N 1, N 2 ∈ N such that

∣y∣ ε
∣x n - x ∣< for n ≥ N 1 and
2
∣y∣2 ε
∣y n - y ∣< for n ≥ N 2.
4M
Let N = max { N 0, N 1, N 2 }. Then, n ≥ N implies

∣ xy n

n
-
x
y ∣ =∣ xn
yn
-
xn
y
+
xn
y
-
x
y ∣
≤ ∣x n ∣ ∣ 1
yn
-
1
y ∣ + ∣ 1y ∣∣x n - x∣

∣x n ∣
=
∣y n y∣
∣y - y n∣ + ∣ 1y ∣∣x - x∣
n

2 M ⋅ ∣y∣2 ε
+ ∣ ∣⋅
1 ∣y∣ ε
<
∣y∣2 4M y 2
ε ε
= + = ε
2 2

4. Suppose that x ∈ R, x n ≥ 0, and xn → x as n → ∞. Prove that xn → x

These solutions might be inexact, yet do you rely on them?

- 10 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
as n → ∞. (For the case x = 0 you may wish to use (8) on p.7.)
Proof) Since x n ≥ 0 for all n ∈ N, it is clear that x ≥ 0.
(i ) Case 1 : x = 0.
Since x n → 0 as n → ∞.

Then, for given ε > 0, we can choose N ∈ N such that

n ≥ N implies ∣x n - 0 ∣< ε 2.
Then, if n ≥ N, then

∣ xn - x ∣= ∣ x n - 0∣< ε 2 = ε.
(ii ) Case 2 : x > 0.
Since x n → x as n → ∞, there is an N ∈ N such that
n ≥ N implies ∣x n - x ∣< x ε.
Then, for given ε > 0, if n ≥ N, then
∣x n - x ∣
∣ xn - x ∣ =
∣ xn + x ∣
∣x n - x ∣
=
xn + x
∣x n - x ∣
<
x
x ε
< = ε
x

5. Prove that given x ∈ R there is a sequence r n ∈ Q such that r n → x as

n → ∞.
Proof) By 1.3Exercises 4, there is a rational r n such that
1
∣x - r n∣< for each n ∈ N.
n
1
By Archimedean Principle, there is an N ∈ N such that N > .
ε
Thus, for given ε > 0, n ≥ N implies
1 1
∣x - r n∣< ≤ < ε.
n N
Hence, r n → x as n → ∞.

<NOTE> A proof of 1.3Exercises 4.

Prove that for each a ∈ R and each n ∈ N there exists a rational r n such that
1
∣a - r n∣< .
n

These solutions might be inexact, yet do you rely on them?

- 11 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
a + rn
Proof) Let r n + 1 = be a rational.
2
For n = 1, by Density of rationals, if a - 1, a + 1 ∈ R,
then ∃ r 1 ∈ Q such that a - 1 < r 1 < a + 1. That is, ∣a - r 1∣< 1.

For some n ∈ N,
1
suppose that ∃ r n ∈ Q such that ∣a - r n∣< . Then,
n

∣a - r n + 1∣ = ∣a - a + rn
2 ∣= ∣ a - rn
2 ∣< 1
2n
<
1
n+1
.

1
That is, we can choose r n + 1 ∈ Q such that ∣a - r n + 1∣ < .
n+1
1
Hence, for each n ∈ N, it holds that ∣a - r n∣ < .
n

6. Suppose that x and y are extended real numbers and { xn }, { yn } and { wn }

are real sequences.
(a) If x n → x and yn → x, as n → ∞, and x n ≤ w n ≤ y n for n ∈ N, prove
that wn → x as n → ∞.
Proof) Case 1. x = ∞.
Since x n → ∞ as n → ∞, by Definition 2.14,

for given M ∈ R we can choose an N ∈ N such that

n ≥ N implies xn > M.
Thus wn ≥ xn > M for all n ≥ N.
Hence wn → ∞ as n → ∞.
Case 2. x = -∞.
Since y n → -∞ as n → ∞, by Definition 2.14,

for given M ∈ R we can choose an N ∈ N such that

n ≥ N implies xn < M.
Thus wn ≤ yn < M for all n ≥ N.
Hence w n → -∞ as n → ∞.

Case 3. -∞ < x < ∞.

By Squeeze Theorem, it is clear.
(b) If x n → x and y n → y, as n → ∞, and x n ≤ y n for n ∈ N, prove that
x ≤ y.
Proof) See Theorem 2.17 (p.43).

7. Using the result in Exercise 4, show the following.

These solutions might be inexact, yet do you rely on them?

- 12 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
(a) Suppose that x 1 ≥ 0 and xn + 1 = 2 + xn for n ∈ N. If xn → x as
n → ∞, prove that x = 2.
Showing) By 2.2Exercises 4, if x n → x as n → ∞, then

xn → x as n → ∞.
Similarly, since x n + 2 → x + 2 as n → ∞, then

xn + 2 → x + 2 as n → ∞.

Thus, for given x n + 1 = 2 + xn ,

x n +1 → x and 2 + xn → x + 2 , as n → ∞.

That is, we conclude by Theorem 2.17 that x = 2+x .

And, x 2 = 2 + x, and ( x + 1 ) ( x - 2 ) = 0.
For all n ∈ N, since x n ≥ 0, then x =
/ -1.
Hence, x = 2.
(b) Suppose that 0 ≤ x 1 ≤ 1 and x n + 1 = 1 - 1 - x n for n ∈ N. If xn → x as
n → ∞, prove that x = 0 or 1.
Showing) By 2.2Exercises 4, if x n → x as n → ∞, then

xn → x as n → ∞.
Similarly, since 1 - x n → 1 - x as n → ∞, then

1 - xn → 1 - x as n → ∞.

Thus, for given x n + 1 = 1 - 1 - xn ,

x n +1 → x and 1 - 1 - xn → 1- 1 - x , as n → ∞.

That is, we conclude by Theorem 2.17 that x = 1 - 1-x .

2
And, ( x - 1 ) = 1 - x, and x ( x - 1 ) = 0.
Hence, x = 0 or 1.

8. Prove Corollary 2.16.

Corollary 2.16 Let { xn }, { yn } be real sequences and α, x, y be extended real

numbers. If xn → x and yn → y, as n → ∞, then lim ( x n + y n ) = x + y

n →∞

(provided that the right side is not of the form ∞ - ∞), and lim ( α x n ) = α x,
n →∞

lim ( x n y n )= x y (provided that none of these products is of the form 0⋅±∞).

n →∞

Proof) If { xn } and { yn } are convergent,

by Theorem 2.12, Corollary 2.16 holds.
On the other hand,
if one and only one of { xn } and { yn } goes to infinity as n → ∞,

These solutions might be inexact, yet do you rely on them?

- 13 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
by Theorem 2.15, Corollary 2.16 holds, also.
Thus, we consider that both { xn } and { yn } go to infinity as n → ∞.
(i ) lim ( x n + y n ) = x + y.
n →∞

It suffices to show that lim ( x n + y n ) = ∞ + ∞ = ∞.

n →∞

If both { xn } and { yn } go to infinity as n → ∞,

let M ∈ R, we can choose N 1, N 2 ∈ N

such that n ≥ N 1 and n ≥ N 2 imply

M M
xn > and yn > .
2 2
Set N = max { N 1, N 2 }, then n ≥ N implies
M M
xn + y n > + = M.
2 2
Similarly, if x n → -∞ and y n → -∞, as n → ∞, then
lim ( x n + y n ) = (-∞ ) + ( -∞ ) = -∞.
n→∞

(ii ) lim ( α x n )= α x.
n →∞

By Trichotomy Property and 1.1Exercises 11 (a), since we consider

provided that none of these products is of the form 0⋅±∞, then
α > 0 and α < 0.
By Theorem 2.15, it goes to prove that
lim ( α x n )= α x holds for α > 0.
n →∞

Suppose that α < 0. Then, -α > 0.

It suffices to show that lim ( α x n ) = α ⋅∞ = -∞.
n→∞

M
Let M ∈ R and M0 = > 0. We can choose N ∈ N such that
-α
n ≥ N implies xn > M0.
Then, n ≥ N implies
M
α x n < α M0 = α ⋅ by the second Multiplicative Property.
-α
That is, α x n < -M.
Hence, lim ( α x n ) = -∞.
n →∞

Similarly, if x n → -∞ as n → ∞, then lim ( α x n ) = ∞.

n →∞

(iii ) lim ( x n y n )= x y.
n →∞

Since both { xn } and { yn } go to infinity as n → ∞,

Let M 1, M 2 ∈ R and M1, M2 > 0.

These solutions might be inexact, yet do you rely on them?

- 14 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
We can choose N 1, N 2 ∈ N such that

n ≥ N 1 (respectively, n ≥ N 2) implies xn > M (respectively, yn > M ).

Set N = max { N 1, N 2 }.

Then, n ≥ N implies xn yn > M M = M.

Similarly, if x n → -∞ and y n → +∞, as n → ∞,
then lim ( x n y n ) = -∞.
n →∞

And, if x n → +∞ and y n → -∞, as n → ∞,

then lim ( x n y n ) = -∞.
n →∞

And, if x n → -∞ and y n → -∞, as n → ∞,

then lim ( x n y n ) = +∞.
n →∞

9. Interpret a decimal expansion 0.a 1 a 2 ⋯ as

n ak
0.a 1 a 2 ⋯ = lim ∑ .
n → ∞ k=1 10 k
Prove that (a) 0.5 = 0.4999 ⋯ and (b) 1 = 0.999 ⋯.
(a) 0.5 = 0.4999 ⋯.
Proof) By interpretation, since a 1 = 4 and a k = 9 for k ≥ 2,
n ak
0.4999 ⋯ = lim ∑
n → ∞ k =1 10 k

( a1 ak
)
n
= lim +∑ k
n→∞ 10 k = 2 10

( )
n
4 9
= lim +∑ k
n→∞ 10 k = 2 10

= lim 

 4
 +9×
1
10 2 1-
10
1
n -1 ( ) 
n→∞   10 1 
1-
 10 
= lim
n→∞
(4
10
+
1
10
1-
1
(
10 n - 1 ))
5
= = 0.5
10
(b) 1 = 0.999 ⋯.
Proof) By interpretation, since a k = 9 for all k ∈ N,

These solutions might be inexact, yet do you rely on them?

- 15 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
ak n
0.999 ⋯ = lim ∑ k
n → ∞ k =1 10
n
9
= lim ∑
n → ∞ k =1 10 k

= lim 9×
1
10
1- (
1
10 n )
n →∞ 1
1-
10
= lim
n →∞
(1-
1
10 n )
= 1

10. This exercise was used in Section 1.4.

1
(a) Suppose that 0 ≤y < for some integer n ≥ 0. Prove that there is an
10n
integer 0 ≤ w ≤ 9 such that
w ≤y <
w 1
+ .
10 n + 1 10 n + 1 10 n + 1
1
Proof) Suppose that 0 ≤ y < for some integer n ≥ 0.
10n
By 2.2Exercises 9, let y = 0.a 1 a 2 ⋯ a n a n + 1 ⋯, then

ak = { 0,

a k ∈ { 0, 1, ⋯ , 9 },
1 ≤k ≤n

k > n
t ak
holds. And since 0.a 1 a 2 ⋯ = lim ∑ , then
t → ∞ k=1 10 k
t ak
y = 0.a 1 a 2 ⋯ a n a n + 1 ⋯ = lim ∑
t → ∞ k=1 10 k
t ak
= lim ∑
t → ∞ k = n +1 10 k
Since a k ∈ { 0, 1, ⋯ , 9 }, then
there is an integer 0 ≤ w ≤ 9 such that w = a n + 1.
t ak
y = lim ∑
t → ∞ k = n +1 10 k

= lim
t→∞
( an+1
10 n +1 +
a n +2
10 n +2 +
a n +3
10 n +3 +
⋯ +
at
10 t )
an+1 w
≥ =
10 n + 1 10 n + 1
and

These solutions might be inexact, yet do you rely on them?

- 16 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
t ak
y = lim ∑
t → ∞ k = n +1 10 k

( an +1 ak
)
t
= lim n +1 + ∑ k
t →∞ 10 k = n +2 10

( a n +1
)
t
9
< lim n +1 + ∑ k
t →∞ 10 k = n +2 10

= lim 


 an +1
+ 9×
10
1
n+2 1 - ( 10
1
t - n -1 ) 
t →∞  10 n + 1 1 
 1- 
 10 
= lim
t →∞
( an +1
10 n + 1
+
1
10 n +1 ( 1-
1
10 t - n -1 ))
w 1
= +
10 n + 1 10 n + 1
w ≤y < w 1
hence, + .
10 n + 1 10 n + 1 10 n + 1
(b) Prove that given x ∈ [ 0, 1 ) there exist integers 0 ≤ x k ≤ 9 such that for all

n ∈ N,
n xk n xk 1
∑ k
≤x < ∑ k + .
k =1 10 k =1 10 10 n
t xk
Proof) Since x ∈ [ 0, 1 ), set x = 0.x 1 x 2 ⋯ = lim ∑ .
t → ∞ k =1 10 k
Then, for 1 ≤ n < t,
xk
t
x = lim ∑ k
t → ∞ k =1 10

( xk xk
)
n t
= lim ∑ k + ∑ k
t →∞ k =1 10 k = n +1 10

n xk
≥ lim ∑ k
t → ∞ k =1 10

n xk
= ∑ k
k =1 10

and

These solutions might be inexact, yet do you rely on them?

- 17 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
xk t
x = lim ∑ k
t → ∞ k =1 10

(x xk
)
n t
= lim ∑ kk + ∑ k
t →∞ k =1 10 k = n +1 10

(
xk
)
n t
9
< lim ∑ k + ∑ k
t →∞ k =1 10 k = n +1 10


 n x k

= lim  ∑
10
1
n +1 1-
10
1
(
t -n



)
k +9
× 
t →∞   k =1 10 1 
1-
 10 

(xk
))
n
= lim
t →∞
∑
k =1 10
k +
1
10 n
1-
1
10 t - n (
n xk 1
= ∑ +
k =1 10 k 10 n
n xk n xk 1
Hence, ∑ k ≤x < ∑ k + .
k =1 10 k =1 10 10 n
(c) Prove that given x ∈ [ 0, 1 ) there exist integers 0 ≤ x k ≤ 9, k ∈ N, such
that
n xk
x = lim ∑ .
n → ∞ k =1 10 k
n xk n xk 1
Proof) By part (b), ∑ k ≤x < ∑ k + .
k =1 10 k =1 10 10 n
n xk 1
That is, 0 ≤ x - ∑ < .
k =1 10 k 10 n
1
Since lim = 0, by Squeeze Theorem, then
n →∞ 10 n

( xk
) = 0.
n
lim x -∑
n →∞ k =1 10 k
n xk
Hence, x = lim ∑ .
n → ∞ k=1 10 k

2.3 Bolzano-Weierstrass Theorem

p.48
Exercises
1. Prove that
( n 2 + 20 n + 35 ) sin ( n 3 )
xn =
n2 + n + 1
has a convergent subsequence.

These solutions might be inexact, yet do you rely on them?

- 18 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Proof) Since -1 ≤ sin ( n 3 ) ≤ 1 for all n ∈ N,
n 2 + 20 n + 35 ≤ ( n 2 + 20 n + 35 ) sin ( n 3 ) ≤ n 2 + 20 n + 35
- .
n2 + n + 1 n2 + n + 1 n2 + n + 1
That is,
n 2 + 20 n + 35 ≤ ≤ n 2 + 20 n + 35
- xn for all n ∈ N.
n2 + n + 1 n2 + n + 1

Since { n 2 + 20 n + 35
n2 + n + 1 } is convergent (goes to 1), by Theorem 2.8,

∃ M > 0 such that

n 2 + 20 n + 35 ≤
M for all n ∈ N.
n2 + n + 1
Thus, ∣x n∣≤ M for all n ∈ N.
Hence, { xn } is bounded. By Bolzano-Weierstrass Theorem,
{ xn } has a convergent subsequence.

2. Suppose that E ⊂ R is a nonempty bounded set and sup E ∉ E. Prove that

there exists a strictly increasing sequence { xn } that converges to sup E such that

x n ∈ E for all n ∈ N.
Proof) Since x n ∈ E for all n ∈ N and E is bounded, { xn } is also bounded.
And since { xn } is strictly increasing, by Monotone Convergence Theorem,
{ xn } converges.
Then, it suffices to prove that there exists { xn } that converges to sup E.

Since sup E ∉ E, then x n =

/ sup E for all n ∈ N.
Thus, by Approximation Property for Suprema, for given ε > 0,
there is an N ∈ N such that sup E - ε < x N < sup E.
Since { xn } is strictly increasing, x N ≤ x n for n ≥ N.

Thus, sup E - ε < x n < sup E for all n ≥ N.

That is, ∣ sup E - x n ∣ < ε for all n ≥ N.
Hence, { xn } converges to sup E.

3. (a) Suppose that { xn } is a monotone increasing sequence in R (not necessarily

bounded above). Prove that there is an extended real number x such that xn → x
as n → ∞.
Proof) If { xn } is bounded above, by Monotone Convergence Theorem,

∃ x ∈ R such that xn → x as n → ∞.
Otherwise, { xn } is not bounded,
we can assume that x n → +∞ as n → ∞.

These solutions might be inexact, yet do you rely on them?

- 19 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Since { xn } is not bounded,

for each M ∈ R, there is an N ∈ N such that

xn > M for infinitely many n ≥ N.
But, since { xn } is a monotone increasing sequence,
for all n ≥ N, it holds that x n ≥ xN > M.

Hence, x n → +∞ as n → ∞.
(b) State and prove an analogous result for decreasing sequence.
Proof) Omission!

4. Suppose that 0 < x1 < 1 and xn+1 = 1 - 1 - xn for n ∈ N. Prove that

x n +1 1
x n ↓ 0 as n → ∞ and → , as n → ∞. (Exercise 4.3 in Apostol [1].)
xn 2

Proof) First, we need to show that 0 < xn < 1 for all n ∈ N.

Suppose that 0 < xn < 1.

Then, 0 < 1 - xn < 1 and 0 < 1 - xn < 1.

And, 0 < 1 - 1 - xn < 1. That is, 0 < xn +1 < 1.

Thus, by induction, 0 < xn < 1 for all n ∈ N.

Second, we consider that
2
( 1 - x n ) 2 - ( 1 - x n +1 ) 2 = ( 1 - x n ) 2 - ( 1 - xn )
= ( 1 - xn )2 - ( 1 - xn )

= xn ( xn - 1 )

< 0
where 0 < xn < 1 for all n ∈ N.
Then, since 1 - x n > 0 and 1 - x n +1 > 0, it follows that

1 - x n < 1 - x n + 1. That is, x n > x n +1 for all n ∈ N.

Thus, since { xn } is decreasing and bounded below by 0,
by Monotone Convergence Theorem, { xn } converges.
Third, we need to show that lim x n = 0.
n →∞

Suppose that L : = lim x n for some L ∈ R.

n→∞

Since L = lim x n = lim x n + 1,

n →∞ n →∞

taking the limit of an equality x n + 1 = 1 - 1 - xn ,

we have L = 1 - 1-L .
Then, L = 0 or L = 1.

These solutions might be inexact, yet do you rely on them?

- 20 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
If L = 1, since x1 < 1, it contradicts that x n > x n +1 for all n ∈ N.
Thus, L = 0. And, since { xn } is decreasing,
thus, x n ↓ 0 as n → ∞.

x n +1 1 - 1 - xn
On the other hand, we have = . And,
xn xn

x n +1 1 - 1 - xn
=
xn xn
( 1 - 1 - xn ) ( 1 + 1 - xn )
=
x n ( 1 + 1 - xn )
1 - ( 1 - xn )
=
x n ( 1 + 1 - xn )
1
=
1 + 1 - xn
x n +1 1
Since lim x n = 0, thus → as n → ∞.
n →∞ xn 2

5. Let 0 < x1 ≤ 3 and xn +1 = 2xn + 3 for n ∈ N. Prove that x n ↑ 3 as

n → ∞.
Proof) First, we need to show that 0 < xn < 3 for all n ∈ N.
Suppose that 0 < xn < 3.
Then, 0 < 2xn < 6 and 3 < 2 xn + 3 < 9.

And, 3 < 2xn +3 < 3. That is, 0 < xn +1 < 3.

Thus, by induction, 0 < xn < 3 for all n ∈ N.

Second, we consider that
x n2+1 - x n2 = 2 x n + 3 - x n2

= -( x n + 1 ) ( x n - 3 )

> 0
where 0 < xn < 3 for all n ∈ N.
Then, since xn > 0 and x n +1 > 0, it follows that

x n +1 > x n for all n ∈ N.

Thus, since { xn } is increasing and bounded above by 3,
by Monotone Convergence Theorem, { xn } converges.
Third, we need to show that lim x n = 3.
n →∞

Suppose that L : = lim x n for some L ∈ R.

n→∞

These solutions might be inexact, yet do you rely on them?

- 21 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Since L = lim x n = lim x n + 1,
n →∞ n →∞

taking the limit of an equality x n + 1 = 2xn + 3 ,

we have L = 2L + 3 .
Then, L = -1 or L = 3.
If L = -1, it contradicts that 0 < xn < 3 for all n ∈ N.
Thus, L = 3. And, since { xn } is increasing,
thus, x n ↑ 3 as n → ∞.

6. Suppose that x 1 ≥ 2 and x n + 1 = 1 + x n - 1 for n ∈ N. Prove that xn ↓ 2

as n → ∞. What happens when 1 ≤ x1 < 2?

Proof) First, we need to show that x n ≥ 2 for all n ∈ N.

Suppose that x n ≥ 2.

Then, x n - 1 ≥ 1 and x n - 1 ≥ 1.

And, 1 + x n - 1 ≥ 2. That is, x n + 1 ≥ 2.

Thus, by induction, x n ≥ 2 for all n ∈ N.

Second, we consider that
( x n + 1 - 1 ) 2 - ( xn - 1 )2 = ( xn - 1 ) - ( xn - 1 ) 2

= ( xn - 1 ) - ( xn - 1 )2

= -( x n - 2 ) ( x n - 1 )

≤0

where x n ≥ 2 for all n ∈ N.

Then, since xn - 1 > 0 and x n +1 - 1 > 0, it follows that

x n + 1 - 1 ≤ x n - 1. That is, x n + 1 ≤ x n for all n ∈ N.

Thus, since { xn } is decreasing and bounded below by 2,
by Monotone Convergence Theorem, { xn } converges.
Third, we need to show that lim x n = 2.
n →∞

Suppose that L : = lim x n for some L ∈ R.

n→∞

Since L = lim x n = lim x n + 1,

n →∞ n →∞

taking the limit of an equality x n + 1 = 1 + xn - 1 ,

we have L = 1 + L-1 .
Then, L = 1 or L = 2.
If L = 1, it contradicts that x n ≥ 2 for all n ∈ N.

These solutions might be inexact, yet do you rely on them?

- 22 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Thus, L = 2. And, since { xn } is decreasing,
thus, x n ↓ 2 as n → ∞.
Similarly, if 1 ≤ x1 < 2,

1 ≤ xn < 2 and x n + 1 ≥ x n for all n ∈ N.

Thus, x n ↑ 2 as n → ∞.

7. Prove that
 1 x>0

1
2n-1

lim x =  0 x=0
n →∞ 

 -1 x < 0.
Proof) ① Case 1 : x > 0.
There are three cases.
1 1
2n -1 2n -1
If x = 1, then 1 = 1 for all n ∈ N. Thus x → 1 as n → ∞.
If x > 1, for fix n ∈ N, x > 1 implies x 2n +1 > x 2n -1 > 1.
Then, taking the ( 2 n - 1 ) ( 2 n + 1 )st root of this inequality,
1 1 1
2n -1 2n +1 2n -1
x > x > 1. That is, x is decreasing and bounded below.
1
2n -1
By Monotone Convergence Theorem, L : = lim x for some L ∈ R.
n→∞

1 1 1 1 1
Then, L = lim x
n →∞
2n -1
= lim x
n →∞
2n
= lim
n →∞
(x )
n 2
= L 2
.

1
2n -1
That is, L = L 2. Then L = 0 or L = 1. Since x > 1,
hence L = 1.
1
Otherwise, if 0 < x < 1, then > 1. By Theorem 2.12 and case of x > 1,
x
1
2n -1 1 1
lim x = lim 1 = 1 = 1.
( ) 1
( )1
n →∞ n →∞ 2n -1 2n -1
lim
x n →∞ x
② Case 2 : x = 0.
1 1 1
2n -1 2n -1 2n -1
Since x = 0 = 0, then x converges to 0 immediately.
③ Case 3 : x < 0.
2n - 1
Since 2 n - 1 is odd integer for all n ∈ N, then ( -1 ) = -1.
1
2n -1
That is, ( -1 ) = -1.
If x < 0, then -x > 0. Thus, by case of ①,

These solutions might be inexact, yet do you rely on them?

- 23 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
1 1 1 1
2n -1 2n -1 2n -1 2n -1
lim x = lim (-1 ) (- x ) = - lim (- x ) = -1.
n →∞ n →∞ n →∞

1 + xn - 1
8. Suppose that x 0 ∈ R and x n = for n ∈ N. Prove that xn → 1 as
2
n → ∞.
Strategy) Geometrically, there are three cases.
1 +1
First, if x 0 = 1, since x 1 = = 1,
2
we assume that x n = 1 for all n ∈ N.
Second, if x0 > 1, since x1 is the arithmetic mean of 1 and x 0,
(in generally, x n is the arithmetic mean of 1 and x n - 1)
{ xn } is decreasing and bounded below by 1.
Third, similarly, if x0 < 1,
{ xn } is increasing and bounded above by 1.
By Monotone Convergence Theorem,
{ xn } has a finite limit which equals to 1.
Proof) ① Case 1. x 0 = 1.

For all n ∈ N, x n = 1. Thus, xn → 1 as n → ∞.

② Case 2. x 0 =
/ 1.
1 + xn -1
For given x n = , then
2
1 + xn -1 xn -1 - 1
xn - 1 = -1 = . Thus, we assume that
2 2
x n -1 - 1 xn -2 - 1 x0 - 1
xn - 1 = = = ⋯⋯ = .
2 22 2n
x0 - 1
That is, we assume that x n - 1 = .
2n
x0 - 1
Then, we need to prove that x n - 1 = .
2n
x0 - 1
For n = 1, it is obvious that x 1 - 1 = .
21
x0 - 1
Suppose that x n - 1 = . Then
2n
1 + xn xn - 1 1 ⋅ x0 - 1 x0 - 1
xn + 1 - 1 = -1 = = = .
2 2 2 2n 2n + 1
x0 - 1
That is, x n + 1 - 1 = . By induction,
2n +1

These solutions might be inexact, yet do you rely on them?

- 24 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
we conclude that the formula holds for all n ∈ N.
∣x 0 - 1∣
Let ε > 0 be given. For N > 0 and > 0,
ε
by Archimedean Principle, ∃ N ∈ N such that
∣x 0 - 1∣
N > .
ε
Then, if n ≥ N, then
∣x 0 - 1∣ ∣x 0 - 1∣ ∣x 0 - 1∣
∣x n - 1∣= ≤ < < ε.
2n 2N N
Hence, x n → 1 as n → ∞.

9. Let 0 < y1 < x1 and set

xn + yn
xn +1 = and y n + 1 = x n y n , n ∈ N.
2
(a) Prove that 0 < yn < xn for all n ∈ N.
Proof) The proof is by induction on n.
For n = 1, by hypothesis, it's clear!
Suppose that 0 < yn < xn.

Then, by Example 1.3 (7), (8), y n +1 = x n y n > 0.

And, by Example 1.2, since x n =
/ y n, then
x n + yn
x n + 1 - yn + 1 = - xn yn
2
1 2
=
2 ( xn - yn )

> 0
That is, x n +1 > y n +1.

Hence, 0 < yn < xn for all n ∈ N.

(b) Prove that y n is increasing and bounded above, and x n is decreasing and
bounded below.
Proof) Since 0 < yn < xn for all n ∈ N,
yn + 1 - yn = xn yn - yn > y n y n - y n = 0.

That is, y n +1 > y n for all n ∈ N.

Thus, y n is increasing.
xn + yn yn - xn
Similarly, x n + 1 - x n = - xn = < 0.
2 2
That is, x n +1 < x n for all n ∈ N.
Thus, x n is decreasing.

These solutions might be inexact, yet do you rely on them?

- 25 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
On the other hand, since x n is decreasing and 0 < yn < xn for all n ∈ N,

then 0 < yn < xn ≤ x1 for n ≥ 1 and n ∈ N.

Thus, y n is bounded above by x1.

Similarly, since y n is increasing and 0 < yn < xn for all n ∈ N,

then 0 < y1 ≤ yn < xn for n ≥ 1 and n ∈ N.

Thus, x n is bounded below by y 1.
x1 - y1
(c) Prove that 0 < x n +1 - y n +1 < for n ∈ N.
2n
x1 - y1
Proof) It suffices to prove that x n +1 - y n + 1 < .
2n
First, since 0 < yn < xn for all n ∈ N,
then xn , yn > 0 and xn - yn > 0. Thus,

xn - yn = ( xn - yn )( xn + yn )

> ( xn - yn ) ( xn - yn )
2
= ( xn - yn )
2
That is, ( xn - yn ) < xn - yn.
For n = 1,
x1 + y1
x2 - y2 = - x1 y1
2
1 2
=
2 ( x1 - y1 )
x1 - y1
<
21
x1 - y1
Suppose that 0 < x n +1 - y n +1 < . Then,
2n
xn +1 + yn +1
xn +2 - yn +2 = - xn +1 yn +1
2
1 2
=
2 ( xn +1 - yn +1 )

1
< ( xn +1 - yn +1 )
2

<
1
2 ( x1 - y1
2n )
x1 - y1
=
2n +1
We conclude by induction that the formula holds for all n ∈ N.
(d) Prove that lim x n = lim y n. (This common value is called the
n →∞ n →∞

These solutions might be inexact, yet do you rely on them?

- 26 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
arithmetic-geometric mean of x 1 and y1.)
Proof) By part (b), y n is increasing and bounded above,
and x n is decreasing and bounded below.
Thus, by Monotone Convergence Theorem, y n and x n are convergent.
By Theorem 2.12 and part (c),
lim x n - lim y n = lim ( x n - y n ) and
n →∞ n →∞ n →∞

x1 - y1
lim 0 ≤ lim ( x n - y n ) ≤ lim = 0.
n →∞ n→∞ n →∞ 2n - 1
Thus, by Squeeze Theorem, lim x n - lim y n = 0.
n→∞ n →∞

10. Suppose that x 0 = 1, y 0 = 0,

x n = x n - 1 + 2 y n - 1,
and
yn = xn -1 + yn -1

for n ∈ N. Prove that x n2 - 2 y n2 = ± 1 for all n ∈ N and

xn
→ 2 as n → ∞.
yn

Proof) First, prove that x n2 - 2 y n2 = ± 1 for all n ∈ N.

Since x n = x n - 1 + 2 y n - 1 and y n = x n - 1 + y n - 1, then

x n2 - 2 y n2 = ( x n - 1 + 2 y n - 1 ) 2 - 2 ( x n - 1 + y n - 1 ) 2

= -x n2-1 + 2 y n2-1

= (-1 ) ( x n2-1 - 2 y n2-1 ) 2

We assume that
x n2 - 2 y n2 = (-1 ) ( x n2-1 - 2 y n2-1 ) 2

= (-1 ) 2 ( x n2-2 - 2 y n2-2 ) 2

⋯⋯⋯

= (-1 ) n ( x 02 - 2 y 02 ) 2

= (-1 ) n
Thus, x n2 - 2 y n2 = -1 for all odd n ∈ N,

x n2 - 2 y n2 = 1 for all even n ∈ N.

The proof is by induction on n.
For n = 1, x 12 - 2 y 12 = (-1 ) ( x 02 - 2 y 02 ) 2 = -1. (Clear!)

These solutions might be inexact, yet do you rely on them?

- 27 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Suppose that x n2 - 2 y n2 = (-1 ) n. Then,

x n2+1 - 2 y n2+1 = (-1 ) ( x n2 - 2 y n2 ) 2 = (-1 ) (-1 ) n = (-1 ) n +1.

We conclude by induction that the formula holds for all n ∈ N.

xn
Second, prove that → 2 as n → ∞.
yn
xn
To prove → 2,
yn

we need to prove that x n , y n ≥ 1 and y n ≥ n for all n ∈ N.

For n = 1, it is clear that
x 1 = x 0 + 2 y 0 = 1 + 2 = 3 ≥ 1 and
y 1 = x 0 + y 0 = 1 + 0 = 1 ≥ 1.
Suppose that x n , y n ≥ 1 and y n ≥ n hold for some n.
Then,
x n + 1 = x n + 2 y n ≥ 3 ≥ 1 and
y n + 1 = x n + y n ≥ 2 ≥ 1 and
y n + 1 = x n + y n ≥ n +1.

We conclude by induction that x n , y n ≥ 1 and y n ≥ n for all n ∈ N.

Let ε > 0, by Archimedean Principle,
1
there is an N ∈ N such that N > . If n ≥ N, then
ε

∣ xy n

n
- 2 ∣ =∣ x - y 2 y ∣
n

n
n

= ∣ x n2 - 2 y n2
y n ( x n + 2 yn ) ∣
=∣ ±1
y n ( x n + 2 yn ) ∣
≤∣ 1
∣
y (1+n 2 )

< ∣
y ∣
1
n

≤ 1
n
xn
By Squeeze Theorem, → 2 as n → ∞.
yn

Reference :
<NOTE> A proof of the Geometric-Harmonic Mean
[the Geometric-Harmonic Mean] Suppose that x0 > 0, y0 > 0 and x 0 ≤ y 0,

These solutions might be inexact, yet do you rely on them?

- 28 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
2 xn -1 yn -1
xn = , and
xn -1 + yn -1

yn = x n - 1 y n - 1 for all n ∈ N.
(This common value is called the geometric-harmonic mean of x n - 1 and y n - 1.
That is, x n is called the harmonic mean of x n - 1 and y n - 1,
and y n is called the geometric mean of x n - 1 and y n - 1 .)

(a) Prove that x n ↑ x and y n ↓ y, as n → ∞, for some x, y ∈ R. And prove

that x n - 1 ≤ x n ≤ y n -1 and x n - 1 ≤ y n ≤ y n - 1.

Proof) Suppose that x n , y n > 0 for some n ∈ N ∪ { 0 }.

2xn yn
Then, x n + 1 = and y n + 1 = x n y n are also positive.
xn + yn

( )
2
1 1 2 1 1
Thus, we see + - = - ≥ 0.
xn -1 yn -1 xn -1 yn -1 xn -1 yn -1
And, we have
1 1 2 xn -1 + yn -1 2
+ ≥ and ≥ .
xn -1 yn -1 xn -1 yn -1 xn -1 yn -1 xn -1 yn -1
2xn -1 yn -1
That is, ≤ xn -1 yn -1 .
xn -1 + yn -1

This is equivalent to x n ≤ y n for all n ∈ N ∪ { 0 }.

Since x n ≤ y n for all n ∈ N ∪ { 0 },

2xn -1 yn -1 xn -1 ( yn -1 - xn -1 )
xn - xn -1 = - xn -1 = ≥0
xn -1 + yn -1 xn -1 + yn -1

and y n - y n - 1 = x n - 1 y n - 1 - y n - 1 ≤ y n - 1 - y n - 1 = 0.
And, since x0 ≤ y0 and x n ≥ x n - 1, y n ≤ y n - 1, we have
x 0 ≤ x 1 ≤ x 2 ≤ ⋯ ≤ x n ≤ ⋯ ≤ y n ≤ ⋯ ≤ y 2 ≤ y 1 ≤ y 0.
That is, { xn } is increasing and bounded above,
{ yn } is decreasing and bounded below.

By Monotone Convergence Theorem, { xn } and { yn } have finite limits.

(b) Set x = lim x n and y = lim y n. Prove that x = y.
n →∞ n →∞

Proof) Set x = lim x n and y = lim y n.

n →∞ n →∞

2 xn -1 yn -1
Taking the limit of equalities x n = and y n = xn -1 yn -1 .
xn -1 + yn -1
2x y
We have x = and y = xy .
x+y
That is, x ( x - y ) = 0 and y ( y - x ) = 0.

These solutions might be inexact, yet do you rely on them?

- 29 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
Since x0 > 0, y0 > 0, x n ≤ y n and { xn } is increasing, then x, y > 0.
Hence, x = y.

11. [Archimedes] Suppose that x 0 = 2 3 , y 0 = 3,

2 xn -1 yn -1
xn = , and
xn -1 + yn -1

yn = x n y n - 1 for all n ∈ N.

(a) Prove that x n ↓ x and y n ↑ y, as n → ∞, for some x, y ∈ R.

Proof) By NOTE (See a reference above the theorem),
since x n is the harmonic mean of x n - 1 and y n - 1, then

x n lies between x n - 1 and y n - 1 for all n ∈ N.

And, since y n is the geometric mean of x n and y n - 1, then

y n lies between x n and y n - 1 for all n ∈ N.

Since x0 = 2 3 > 3 = y0, then y n -1 < x n < x n -1 and yn -1 < yn < xn.

(If not, a hypothesis of 'for all n ∈ N' can not be strict.)

Thus, x n is strictly decreasing and y n is strictly increasing.

(b) Prove that x = y and

3.14155 < x < 3.14161.
(The actual value of x is π.)
Proof) By part (a), since y n -1 < x n < x n -1 and yn -1 < yn < xn for all n ∈ N,
then x n is bounded below by y 0 and y n is bounded above by x0.
By part (a) and Monotone Convergence Theorem,
x n and y n have finite limits.
Set x = lim x n and y = lim y n.
n →∞ n →∞

2 xn -1 yn -1
Taking the limit of equalities x n = and y n = xn yn -1 .
xn -1 + yn -1
2x y
We have x = and y = xy .
x+y
That is, x ( x - y ) = 0 and y ( y - x ) = 0.
Since x n ≥ y 0 = 3 and y n ≥ y 0 = 3, then x, y > 0.
Thus, x = y.
In addition, by part (a) and (b),
we assume that yn < x = y < xn for all n ∈ N.

These solutions might be inexact, yet do you rely on them?

- 30 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
To estimate the value x, calculate values of x n's and y n's.
For n = 8, x 8 is approximately equal to 3.14159703432153 and
for n = 6, y 6 is approximately equal to 3.14155760791186.

That is, x ∈ R such that

3.14155760791186 ≒ y 6 < x < x 8 ≒ 3.14159703432153.

Hence, 3.14155 < x < 3.14161.

For reference,
the following values are approximate values of x n's and y n's.

(to an accuracy of 10 - 14)

x 1 = 3.21539030917347, y 1 = 3.10582854123025,
x 2 = 3.1596599420975, y 2 = 3.13262861328124,
x 3 = 3.14608621513143, y 3 = 3.13935020304687,
x 4 = 3.14271459964537, y 4 = 3.14103195089051,
x 5 = 3.14187304997982, y 5 = 3.14145247228546,
x 6 = 3.14166274705685, y 6 = 3.14155760791186,
x 7 = 3.14161017660469, y 7 = 3.14158389214832,
x 8 = 3.14159703432153, y 8 = 3.14159046322805,
x 9 = 3.14159374877135, y 9 = 3.14159210599927,
x 10 = 3.1415929273851, y 10 = 3.14159251669216.

2.4 Cauchy Sequences

p.51
Exercises
1. Prove that (without using Theorem 2.29) that the sum of two Cauchy sequence
is Cauchy.
Proof) Let { xn } and { yn } be Cauchy sequences.

Then, for given ε > 0, there are N 1, N 2 ∈ N such that

ε
n, m ≥ N 1 imply ∣x n - x m∣< and
2
ε
n, m ≥ N 2 imply ∣y n - y m∣< .
2
Set N = max { N 1, N 2 }, then n ≥ N implies

These solutions might be inexact, yet do you rely on them?

- 31 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
∣( x n + y n ) - ( x m + y m )∣ = ∣( x n - x m ) + ( y n - y m )∣

≤∣x n - x m∣+∣y n - y m∣

ε ε
< + = ε
2 2
Thus, { x n + y n } is also Cauchy.

Reference :
<NOTE> Prove that (without using Theorem 2.29) that the product of two Cauchy
sequence is Cauchy.
Proof) Let { xn } and { yn } be Cauchy sequences.

Then, for given ε = 1, there are N 1, N 2 ∈ N such that

n, m ≥ N 1 imply ∣x n - x m∣< 1 and
n, m ≥ N 2 imply ∣y n - y m∣< 1.
Here, if you set n = N 3 = max { N 1, N 2 },
then we have ∣x N 3 - x m∣< 1 and ∣y N 3 - y m∣< 1,
that is, by the Triangle Inequality,
∣x m∣ < 1 +∣x N 3∣ and ∣y m∣ < 1 +∣y N 3∣
by the Triangle Inequality.
Thus, { xn } and { yn } are bounded by some M > 0,

that is, ∣x m∣< M and ∣y m∣< M for all m ∈ N.

For any ε > 0, there is an N 0 ∈ N such that
ε
n, m ≥ N 0 imply ∣x n - x m∣< and
2M
ε
n, m ≥ N 0 imply ∣y n - y m∣< .
2M
Finally, if n > N0, then
∣x n y n - x m y m∣ = ∣x n y n - x m y n + x m y n - x m y m∣

≤ ∣x n y n - x m y n∣ + ∣x m y n - x m y m∣

= ∣y n∣∣x n - x m∣ + ∣x m∣∣y n - y m∣

ε ε
< M⋅ +M⋅
2 M 2 M
= ε

2. Prove that if { xn } is a sequence that satisfies

1+n
∣x n∣≤
1 + n + 2n2

These solutions might be inexact, yet do you rely on them?

- 32 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
for all n ∈ N, then { xn } is Cauchy.
Proof) By Squeeze Theorem and Theorem 2.12,
1 1
+
1+n n2 n 0+0
lim = lim = = 0.
n→∞ 1 + n + 2 n2 n →∞ 1
+
1
+2
0+0+2
n2 n
By Squeeze Theorem and since
1+n 1+n
- ≤ xn ≤ ,
1 + n + 2n2 1 + n + 2n2
then lim x n = 0.
n →∞

By Theorem 2.29, { xn } is Cauchy.

3. Suppose that x n ∈ N for all n ∈ N. If { xn } is Cauchy, prove that there are

numbers a and N such that x n = a for all n ≥ N.

Proof) Suppose that { xn } is Cauchy.

1
Then, for given ε = , there is an N ∈ N such that
2
1
n, m ≥ N imply ∣x n - x m∣< .
2
Since x n ∈ N for all n ∈ N, by Remark 1.1 and Remark 1.13,
x n = x m or ∣x n - x m∣≥ 1 for all n, m ≥ N.
If x n = x m for all n, m ≥ N, then x n is a constant a as promised.
Otherwise, ∣x n - x m∣≥ 1 implies
1 1
1 ≤∣x n - x m∣< . That is, 1 < , a contradiction.
2 2
Thus, x n = x m = a for all n, m ≥ N.

4. Let { xn } be a sequence of real numbers. Suppose that for each ε > 0 there is

∣ ∑ x ∣< ε. Prove that

m
an N ∈ N such that m ≥ n ≥ N implies k
k =n

n
lim ∑ xk
n →∞ k =1

exists and is finite.

n
Proof) Set s n : = ∑ x k be a sequence and s 1 = x 1.
k =1

∣ ∣
m
Then, for each ε > 0, ∑ x k < ε implies ∣ s m - s n - 1 ∣< ε.
k=n

That is, { sn } is Cauchy. By Theorem 2.29, { sn } converges.

These solutions might be inexact, yet do you rely on them?

- 33 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
n
Thus, lim ∑ x k = lim s n exists and is finite.
n →∞ k =1 n
→∞

5. Let { xn } be Cauchy. Without using Theorem 2.29, prove that { xn } converges if

and only if at least one of its subsequences converges (compare with Exercise 8,
p. 38).
Proof) (⇒) There are two different proofs of (⇒).
i ) The one way.
Suppose that { xn } is Cauchy and converges.

Given ε = 1, we can choose an N ∈ N such that

∣x n - x N∣< 1 for all n ≥ N. Then, by Triangle Inequalities,

∣x n∣ < 1 +∣x N∣ for all n ≥ N.

Thus, set M : = max { ∣x 1∣, ∣x 2∣, ⋯, ∣x N - 1∣, 1 + ∣ x N∣}, then
{ xn } is bounded by M. By Bolzano-weierstrass Theorem,
{ xn } has a convergent subsequence, say { xnk }.

Hence, at least one { xnk } of its subsequences converges.

ii ) The other way.

Suppose that { xn } converges.

Set n k : = k, then subsequence { xnk } (its own self) of { xn } converges.

Hence, at least one { xnk } of its subsequences converges.

(⇐) Let ε > 0 be given.

Since { xn } is Cauchy, we can choose N 1 ∈ N such that
ε
n, m ≥ N 1 imply ∣x n - x m∣< .
2
And, since subsequence of { xn }, say { xnk }, converges,

for some x ∈ R, we can choose N 2 ∈ N such that

ε
k ≥ N 2 implies ∣x n k - x∣< .
2
Then, for fixed k ≥ N 2 such that n k ≥ N 1,
ε ε
∣x n - x∣≤∣x n - x n k∣ + ∣x n k - x∣ < + = ε for n ≥ N 1.
2 2
Hence, { xn } converges.

n
(-1 ) k
6. Prove that lim ∑ k
exists and is finite.
n →∞ k =1

Proof) Since

These solutions might be inexact, yet do you rely on them?

- 34 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
1 1 1 1 1
- + - + -⋯
n n +1 n +2 n +3 n +4

=
1
n
+ (- 1
n +1
+
1
n +2 )+( - 1
n +3
+
1
n +4 )+⋯
1
< +0+0+⋯
n
1
=
n
and
1 1 1 1 1
- + - + -⋯
n n +1 n +2 n +3 n +4

= ( 1
n
-
1
n +1 )+( 1
n +2
-
1
n +3 )+( 1
n +4
-⋯

> 0+0+0+⋯

= 0
1 1 1 1 1 1
then 0 < - + - + -⋯ < .
n n +1 n +2 n +3 n +4 n
Thus,

-
1
n
< ( -1 ) n ( 1
n
-
1
n +1
+
1
n +2
-
1
n +3
+
1
n +4
-⋯ )< 1
n
for all n ∈ N.
We consider that
(-1 ) k
( )
∞
1 1 1 1 1
∑ = (-1 ) n - + - + -⋯ .
k=n k n n +1 n +2 n +3 n +4
(-1 ) k
∣∑ (-1 ) k
∣<
∞ ∞
1 1 1
Thus, - < ∑ < . That is, .
n k =n k n k =n k n
1
By Archimedean Principle, there is an N ∈ N such that N > .
ε
Then, n ≥ N implies

∣∑ (-1 ) k
∣<
∞
1 ≤ 1
< ε.
k =n k n N

∣ ∑ x ∣< ε as m → ∞,
m
In 2.4Exercises 4, taking the limit of k
k =n

∣ ∑ x ∣< ε implies
∞
we conclude by 2.4Exercises 4 that k
k =n

n
lim ∑ x k exists and is finite.
n →∞ k =1

(-1 ) k n
(-1 ) k
Let x k : =
k
, then lim ∑ k
exists and is finite.
n →∞ k =1

7. Let { xn } be a sequence. Suppose that there is an a > 1 such that

∣x k + 1 - x k∣≤ a - k

These solutions might be inexact, yet do you rely on them?

- 35 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
for all k ∈ N. Prove that xn → x for some x ∈ R.

Proof) Let n, m ≥ N ( n > m ) for n, m ∈ N.

∣x n - x m∣ = ∣x n - x n - 1 + x n - 1 - x n - 2 + ⋯ - x m + 1 + x m + 1 - x m∣

≤ ∣x n - x n - 1∣ + ∣x n - 1 - x n - 2∣+ ⋯ + ∣x m + 1 - x m∣

≤ a-n +1 + a-n +2 + ⋯ + a-m

n-m
= a-n ∑ ak
k=1

a
= a-n ( an -m - 1 )
a-1
a - n + 1⋅a n - m
<
a-1
a1 -m
=
a-1
Taking the limit of the inequality, this implies that
x n - x m → 0 as m → ∞.

That is, { xn } is a Cauchy. Thus, xn → x for some x ∈ R.

8. (a) A subset E of R is said to be sequentially compact if and only if every

sequence x n ∈ E has a convergent subsequence whose limit belongs to E. Prove
that every closed bounded interval is sequentially compact.
Proof) Let E ⊂ R be any closed bounded interval.
Let { xn } be a sequence with x n ∈ E for all n ∈ N.
Since E is a bounded interval, and x n is a bounded sequence,
thus by Bolzano-Weierstrass Theorem,
x n has a convergent subsequence, say x n k.

But suppose that x n k’s limit, say x ∈ R, does not belong to E.

And, since E is closed bounded interval,
there are a, b ∈ R such that E = [ a, b ].
And, set c : = min { ∣x - a∣,∣x - b∣}.
Then, since x ∉ E, x =
/ a and x =
/ b, thus c > 0.
c
Then, for given ε = > 0, there is an N ∈ N such that
2
c
k ≥ N implies ∣x n k - x ∣< .
2
c c
That is, x - < xn k < x + for all k ≥ N.
2 2
It contradicts that x n ∈ E for all n ∈ N.

These solutions might be inexact, yet do you rely on them?

- 36 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim

(b) Prove that there exist bounded intervals in R which are not sequentially
compact.
1
Proof) Set E : = ( 0, 1 ] and x n : = .
n
1
Then, x n ∈ E for all n ∈ N, but lim x n = lim = 0 ∉ E.
n →∞ n →∞ n
(c) Prove that there exist closed intervals in R which are not sequentially
compact.
Proof) Set E : = [ 1, ∞ ) and x n = n.

Then, x n ∈ E for all n ∈ N,

but x n does not have a convergent subsequence.

That is, ∃
/ lim x n.
n →∞

9. (a) Let E be a subset of R. A point a ∈ R is called a cluster point of E if

E ∩ ( a - r, a + r ) contains infinitely many points for every r > 0. Prove that a
is a cluster point of E if and only if for each r > 0, E ∩ ( a - r, a + r ) \ { a } is
nonempty.
Proof) (⇒) Suppose that a is a cluster point of E.
By definition,
E ∩ ( a - r, a + r ) contains infinitely many points for every r > 0.
And, so does E ∩ ( a - r, a + r ) \ { a }.
Thus, E ∩ ( a - r, a + r ) \ { a } is nonempty.
(⇐) For each r > 0, suppose that E ∩ ( a - r, a + r ) \ { a } is nonempty.
Then, this implies that

E ∩ (a- 1
n
, a+
1
n ) \ { a } is nonempty for all n ∈ N.

Since the set is nonempty,

let a n be a real point which belongs to E ∩ (a- 1

n
, a+
1
n ) \ { a }.
1
By Archimedean Principle, there is an N ∈ N such that r > .
N
For all n ≥ N,

since E ∩ (a- 1
n
, a+
1
n ) \ {a }⊆E ∩(a- 1
N
, a+
1
N ) \ { a },
E ∩ (a- 1
N
, a+
1
N ) \ { a } contains infinitely many points.
(For example, the set contains at least a N, a N + 1, a N + 2, ⋯⋯.)
Similarly,

These solutions might be inexact, yet do you rely on them?

- 37 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
since E ∩ ( a-
1
N
, a+
1
N ) \ { a } ⊆ E ∩ ( a - r, a + r ) \ { a },
E ∩ ( a - r, a + r ) \ { a } contains at least a N, a N + 1, a N + 2, ⋯⋯.
And, E ∩ ( a - r, a + r ) contains at least a N, a N + 1, a N + 2, ⋯⋯.
Thus, E ∩ ( a - r, a + r ) contains infinitely many points.
Hence, a is a cluster point of E.
(b) Prove that every bounded infinite subset of R has at least one cluster point.
Proof) Let E be bounded infinite subset of R.
And, let { an } be a sequence with a n ∈ E for all n ∈ N.
Since E is bounded, so is { an }.
By Bolzano-weierstrass Theorem, { an } has a convergent subsequence,

say { an k }, which converges to a for some a ∈ R.

Thus, given ε = r > 0, we can choose an N ∈ N such that

k ≥ N implies ∣a n k - a∣ < ε.

That is, if k ≥ N, then a - r < a n k < a + r.

Since a n k ∈ E, then a n k ∈ E ∩ ( a - r, a + r ) for all k ≥ N.

That is,
E ∩ ( a - r, a + r ) contains infinitely many points for every r > 0.
Hence, a ∈ R is called a cluster point of E.

e
2.5 Limits Supremum and Infimum

p.55
Exercises
1. Find the limit infimum and the limit supremum of each of the following
sequences.
(a) x n = 3 - (-1 ) n.

Solution) Since sup k≥n 3 - (-1 ) n = 4 for all n ∈ N,

by Definition 2.32, limsup n →∞ x n = 4.

Similarly, since inf k ≥n 3 - (-1 ) n = 2 for all n ∈ N,

then liminf n →∞ x n = 2.

These solutions might be inexact, yet do you rely on them?

- 38 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
(b) x n = cos ( nπ
2 ).
Solution) Since sup k ≥n cos ( nπ
2 ) = 1 for all n ∈ N,

by Definition 2.32, limsup n →∞ x n = 1.

Similarly, since inf k ≥n cos ( nπ

2 )= -1 for all n ∈ N,

then liminf n →∞ x n = -1.

(-1 ) n
(c) x n = (-1 ) n + 1 + .
n
1
Solution) For odd n ∈ N, x n = 1 -
n
1
and for even n ∈ N, x n = -1 + .
n
Since sup k ≥n x n = 1 for all n ∈ N,
by Definition 2.32, limsup n →∞ x n = 1.

Similarly, since inf k≥n x n = -1 for all n ∈ N,

then liminf n →∞ x n = -1.

1 + n2
(d) x n = .
2n - 5
Solution) Since
1
2 +1
1 +n n2 0+1 1
lim x n = lim = lim = = ,
n →∞ n→∞ 2n - 5 n→∞ 5 2 -0 2
2-
n
1
by Theorem 2.36, limsup n →∞ x n = liminf n →∞ xn = .
2
yn
(e) x n = , where { yn } is any bounded sequence.
n
Solution) Since { yn } is any bounded sequence, there is an M ∈ R such that
∣y n∣≤ M for all n ∈ N.

Thus, ∣x n∣ = ∣ yn ∣≤ n M
n
for all n ∈ N.

By Squeeze Theorem, xn → 0 as n → ∞.
Hence, by Theorem 2.36, limsup n→∞ xn = liminf n →∞ x n = 0.

(f) x n = n ( 1 + (-1 ) n ) + n - 1 ( (-1 ) n - 1 ).

Solution) Set n k = 2 k -1 for k ∈ N, since

These solutions might be inexact, yet do you rely on them?

- 39 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
x n k = ( 2 k - 1 ) ( 1 + ( -1 ) 2k - 1 ) + ( 2 k - 1 ) - 1 ( (-1 ) 2k - 1 - 1 )

= ( 2 k - 1 ) ( 1 + (-1 )) + ( 2 k - 1 ) - 1 ( (-1 ) - 1 )

-2
=
2k - 1
x n k → 0 as k → ∞.

Otherwise, set n k = 2 k for k ∈ N, since

x n k = 2 k ( 1 + (-1 ) 2k ) + ( 2 k ) - 1 ( (-1 ) 2k - 1 )

= 2k ( 1 + 1 ) + ( 2k )-1 ( 1 - 1 )

= 4k
x n k → ∞ as k → ∞.

Thus, limsup n→∞ x n = ∞ and liminf n →∞ x n = 0.

n3 + n2 - n + 1
(g) x n = .
n 2 + 2n + 5
Solution) By Corollary 2.16,
1 1
n+1- + 2
n3 + n2 - n + 1 n n
lim x n = lim = lim = ∞,
n →∞ n →∞ n 2 + 2n + 5 n →∞
1+
2 5
+ 2
n n
by Theorem 2.36, limsup n →∞ xn = liminf n →∞ x n = ∞.

2. Suppose that { xn } is a real sequence. Prove that

-limsup n→∞ xn = liminf n→∞ (-x n ) and
-liminf n →∞ xn = limsup n→∞ (-x n ).
Proof) By Theorem 1.28 and Definition 2.32,
liminf n →∞ (-x n ) = lim ( inf k ≥n (-x k ) )
n →∞

= lim ( inf { -x k : k ≥ n } )
n →∞

= lim ( -sup { x k : k ≥ n } )
n →∞

= lim ( -sup k ≥n xk )
n →∞

= - lim ( sup k ≥n xk )
n →∞

= -limsup n →∞ xn
and

These solutions might be inexact, yet do you rely on them?

- 40 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
limsup n →∞ (-x n ) = lim ( sup k ≥n (-x k ) )
n →∞

= lim ( sup { -x k : k ≥ n } )
n →∞

= lim ( -inf { x k : k ≥ n } )
n →∞

= lim ( -inf k ≥n xk )
n →∞

= - lim ( inf k ≥n xk )
n →∞

= -liminf n →∞ xn

3. Let { xn } be a real sequence and r ∈ R.

(a) Prove that
limsup n →∞ x n < r implies xn < r
for n large.
Proof) Let s : = limsup n→∞ x n.
Suppose to the contrary that there are n 1 < n 2 < ⋯ such that

x n k ≥ r for all k ∈ N.
Case 1. { xnk } is unbounded above.
By the contraposition of Theorem 2.39,
then s = ∞. Thus, ∞ < r ≤ xn k, a contradiction.

Case 2. { xn k } is bounded above (by some M ∈ R).

Since r ≤ x n k < M, then { xnk } is bounded.

Then, by Bolzano-Weierstrass Theorem and x n k ≥ r,

there is a subsequence of x n that converges at ≥ r.

Thus, s ≥ r, a contradiction.
(b) Prove that
limsup n →∞ x n > r implies xn < r

for infinitely many n ∈ N.

Proof) Let s : = limsup n→∞ x n.

If s > r, then there is an x ∈ R such that r < x ≤ s, and

by Approximation Property for suprema, there is
a subsequence of { xn }, say { xnk }, that converges to x.

That is, x n k > r for large k.

4. Suppose that { xn } and { yn } are real sequences.

(a) Prove that

These solutions might be inexact, yet do you rely on them?

- 41 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
liminf n →∞ x n + liminf n →∞ y n ≤ liminf n →∞ ( xn + yn )

≤ limsup n →∞ x n + liminf n →∞ yn

≤ limsup n →∞ ( xn + yn )

≤ limsup n →∞ x n + limsup n →∞ yn
provided that none of these sums is of the form ∞ - ∞.
Proof) By symmetry and Remark 2.38, it suffices to prove that
liminf n →∞ x n + liminf n →∞ y n ≤ liminf n →∞ ( xn + yn )
and liminf n →∞ ( x n + y n ) ≤ limsup n →∞ x n + liminf n →∞ y n.
By Definition 1.27 (iii ), inf k≥n x k ≤ x k and inf k≥n y k ≤ y k.
Adding these inequalities, we obtain
inf k≥n x k + inf k≥n y k ≤ x k + y k.
Then, { xk + y k } is bounded below by inf k≥n x k + inf k≥n y k.
Thus, by Definition 1.27 (iii ) again,
inf k≥n x k + inf k≥n y k ≤ inf k≥n ( x k + y k ).
Taking the limit of inequality as n → ∞, we have
liminf n →∞ x n + liminf n →∞ y n ≤ liminf n →∞ ( x n + y n ). ⋯⋯ (* )

On the other hand, by (* ) and Remark 2.38,

liminf n →∞ ( x n + y n ) - liminf n →∞ yn

= liminf n →∞ ( x n + y n ) + liminf n →∞ (-y n )

≤ liminf n →∞ ( xn + yn - yn )

= liminf n →∞ xn

≤ limsup n →∞ xn
That is, liminf n →∞ ( x n + y n ) - liminf n →∞ y n ≤ limsup n →∞ xn
and liminf n →∞ ( x n + y n ) ≤ limsup n →∞ x n + liminf n →∞ y n.
(b) Show that if lim x n exists, then
n →∞

liminf n →∞ ( x n + y n ) = lim x n + liminf n →∞ yn

n →∞

and
limsup n→∞ ( x n + y n ) = lim x n + limsup n →∞ y n.
n →∞

Proof) By symmetry, we need only prove that

liminf n →∞ ( x n + y n ) = lim x n + liminf n →∞ y n.
n →∞

By part (a),

These solutions might be inexact, yet do you rely on them?

- 42 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
liminf n →∞ x n + liminf n→∞ y n ≤ liminf n→∞ ( xn + y n )

≤ limsup n→∞ x n + liminf n →∞ yn

And, since lim x n exists, by Theorem 2.36,
n→∞

liminf n →∞ x n = limsup n →∞ x n = lim x n. Thus, we have

n →∞

liminf n →∞ ( x n + y n ) = lim x n + liminf n →∞ y n.

n →∞

(c) Show by examples that each of the inequalities in part (a) can be strict.
Showing) Set x n = (-1 ) n and y n = (-1 ) n + 1. Then,
liminf n →∞ x n + liminf n →∞ y n = (-1 ) + (-1 ) = -2 and
liminf n →∞ ( x n + y n ) = 0 and
limsup n →∞ x n + liminf n →∞ y n = 1 + (-1 ) = 0 and
limsup n →∞ ( x n + y n ) = 0 and
limsup n →∞ x n + limsup n →∞ y n = 1 + 1 = 2. Thus,
liminf n →∞ x n + liminf n→∞ y n < liminf n →∞ ( xn + yn )

= limsup n →∞ x n + liminf n →∞ yn

= limsup n →∞ ( xn + yn )

< limsup n →∞ x n + limsup n →∞ yn

On the other hand, set x n = y n = (-1 ) n. Then,

liminf n →∞ x n + liminf n →∞ y n = (-1 ) + (-1 ) = -2 and
liminf n →∞ ( x n + y n ) = -2 and
limsup n →∞ x n + liminf n →∞ y n = 1 + (-1 ) = 0 and
limsup n →∞ ( x n + y n ) = 2 and
limsup n →∞ x n + limsup n →∞ y n = 1 + 1 = 2. Thus,
liminf n→∞ x n + liminf n→∞ y n = liminf n →∞ ( xn + yn )

< limsup n →∞ x n + liminf n→∞ yn

< limsup n →∞ ( xn + yn )

= limsup n →∞ x n + limsup n →∞ yn
Hence, each of the inequalities in part (a) must be strict.

5. Let { xn } and { yn } be real sequences.

(a) Suppose that x n ≥ 0 and y n ≥ 0 for each n ∈ N. Prove that

limsup n →∞ ( xn yn ) ≤ ( limsup n →∞ xn ) ( limsup n →∞ y n ),

These solutions might be inexact, yet do you rely on them?

- 43 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
provided that the product on the right is not of the form 0⋅∞. Show by example
that this inequality can be strict.
Proof) By Definition 1.16, 0 ≤ x n ≤ sup k≥n x k and 0 ≤ y n ≤ sup k≥n y k.
By Example 1.3, multiplying these inequalities, we obtain
x n y n ≤ ( sup k≥n x k ) ( sup k≥n y k ).
Then, { xn yn } is bounded above by ( sup k≥n x k ) ( sup k≥n y k ).
Thus, by Theorem 2.40,
sup k≥n ( x k y k ) ≤ ( sup k≥n x k ) ( sup k≥n y k ).
Taking the limit of this inequality as n → ∞, we have
limsup n →∞ ( xn yn ) ≤ ( limsup n →∞ xn ) ( limsup n →∞ y n ).

For example, set x n = (-1 ) n and y n = (-1 ) n + 1.

Since x n y n = (-1 ) n (-1 ) n + 1 = (-1 ) 2n + 1 = -1, then

limsup n →∞ ( x n y n ) = -1 and
( limsup n →∞ xn ) ( limsup n →∞ y n ) = 1 × 1 = 1.
Thus, limsup n →∞ ( x n yn ) < ( limsup n →∞ xn ) ( limsup n →∞ y n ).

On the other hand, set x n = y n = (-1 ) n.

limsup n →∞ ( xn yn ) = 1 = ( limsup n →∞ xn ) ( limsup n →∞ y n ).
Hence, this inequality must be strict.
(b) Suppose that x n ≤ 0 ≤ y n for n ∈ N. Prove that
( liminf n →∞ xn ) ( limsup n →∞ yn ) ≤ liminf n →∞ ( x n y n ),

provided that none of these products is of the form 0⋅∞.

Proof 1) By Definition 1.16, inf k≥n xk ≤ xn ≤ 0 and 0 ≤ y n ≤ sup k≥n y k.
By Example 1.3, multiplying these inequalities, we obtain
x n y n ≥ ( inf k≥n xk ) ( sup k≥n y k ).
Then, { xn yn } is bounded below by ( inf k≥n x k ) ( sup k≥n y k ).
Thus, by Theorem 2.40,
inf k≥n ( x k yk ) ≥ ( inf k≥n xk ) ( sup k≥n y k ).
Taking the limit of this inequality as n → ∞, we have
( liminf n →∞ xn ) ( limsup n →∞ yn ) ≤ liminf n →∞ ( xn y n ).
Proof 2) Set w n = - x n ≥ 0.
By part (a),
limsup n →∞ ( wn y n ) ≤ ( limsup n →∞ wn ) ( limsup n →∞ y n ).
By 2.5Exercises 2,

These solutions might be inexact, yet do you rely on them?

- 44 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
limsup n →∞ ( wn y n ) = limsup n →∞ (-x n y n )

= -liminf n →∞ ( xn yn )
( limsup n →∞ wn )( limsup n →∞ yn )

= ( limsup n →∞ (-xn ) )( limsup n →∞ yn )

= -( liminf n →∞ xn )( limsup n →∞ yn )
Hence ( liminf n →∞ xn ) ( limsup n →∞ yn ) ≤ liminf n →∞ ( xn y n ).

6. Suppose that x n ≥ 0 and y n ≥ 0 for all n ∈ N. Prove that if xn → x as

n → ∞ ( x may be an extended real number), then
limsup n →∞ ( xn yn ) = x limsup n →∞ y n,
provided that none pf these products is of the form 0⋅∞.

7. Prove that
limsup n →∞ x n = inf n∈N
( sup k ≥n x k ) and

liminf n →∞ x n = sup n∈ N
( inf k ≥n xk )

for any real sequence { xn }.

Proof) A similar argument establishes the result for the limit infimum.
Let s n = sup k≥n x k be a real sequence.
Case 1. sn is bounded.
By 1.3Exercises 8,
sn = sup k≥n x k is decreasing.
By Monotone Convergence Theorem, { sn } has a finite limit

which satisfies s n → inf { s n : n ∈ N } as n → ∞.

Since lim s n = limsup n →∞ x n,

n →∞

hence limsup n →∞ x n = inf n∈N

( sup k ≥n x k ).

Case 2. s n → ∞ as n → ∞.

1
8. Suppose that x n ≥ 0 for n ∈ N. Under the interpretation = ∞ and
0
1
= 0, prove that
∞
1 1
limsup n →∞ = and
xn liminf n →∞ xn

These solutions might be inexact, yet do you rely on them?

- 45 -  Last Revised : March 30th, 2008.
 The Exercise Solutions of 『An Introduction To ANALYSIS Third Edition』of
William R. Wade, the author of this book.
Written by Gi-uk Kim
1 1
liminf n →∞ = .
xn limsup n → ∞ x n

9. Let x n ∈ R. Prove that x n → 0 as n → ∞ if and only if

lim sup n→∞ ∣x n∣= 0.
Proof) (⇒) Let ε > 0 be given.
Since x n → 0 as n → ∞, there is an N ∈ N such that
n ≥ N implies ∣x n - 0∣< ε.
By the third Triangle Inequality, if n ≥ N, then
∣∣x n∣ - 0∣≤∣x n - 0∣ < ε.
Thus, ∣x n∣ → 0 as n → ∞.
By Theorem 2.36, limsup n→∞ ∣x n∣= 0.
(⇐) Since sup k≥n ∣x k∣≥ 0 and inf k≥n ∣x k∣≥ 0, we have
0 ≤ inf k≥n ∣x k∣≤ sup k≥n ∣x k∣.
By Remark 2.38, taking the limit of these inequalities,
0 ≤ liminf n→∞ ∣x n∣ ≤ limsup n→∞ ∣x n∣.
Since limsup n→∞ ∣x n∣ = 0, we have liminf n →∞ ∣x n∣= 0.
By Theorem 2.36, xn → 0 as n → ∞.

These solutions might be inexact, yet do you rely on them?

- 46 -  Last Revised : March 30th, 2008.